REVIEW CURRENT OPINION A tale of two approaches: how metagenomics and proteomics are shaping the future of encephalitis diagnostics Ryan D. Schubert and Michael R. Wilson Purpose of review We highlight how metagenomics and proteomics-based approaches are being applied to the problem of diagnosis in idiopathic encephalitis. Recent findings Low cost, high-throughput next-generation sequencing platforms have enabled unbiased sequencing of biological samples. Rapid sequence-based computational algorithms then determine the source of all the nonhost (e.g., pathogen-derived) nucleic acids in a sample. This approach recently identified a case of neuroleptospirosis, resulting in a patient’s dramatic clinical improvement with intravenous penicillin. Metagenomics also enabled the discovery of a neuroinvasive astrovirus in several patients. With regard to autoimmune encephalitis, advances in high throughput and efficient phage display of human peptides resulted in the discovery of autoantibodies against tripartite motif family members in a patient with paraneoplastic encephalitis. A complementary assay using ribosomes to display full-length human proteins identified additional autoantibody targets. Summary Metagenomics and proteomics represent promising avenues of research to improve upon the diagnostic yield of current assays for infectious and autoimmune encephalitis, respectively. Keywords 16S rRNA, encephalitis, metagenomics, next-generation sequencing, phage immunoprecipitation sequencing, parallel analysis of translated open reading frame, proteomics INTRODUCTION early administration of acyclovir halved mortality In the United States, there are approximately 20 000 in patients treated promptly [3]. Similarly, Josep cases of encephalitis annually. Inpatient costs alone Dalmau’s discovery of an autoimmune basis for total $2 billion, and the significant morbidity and many causes of encephalitis has revolutionized mortality associated with the condition makes it our understanding of this disease, showing that almost certain that the outpatient and rehabilitative cases presumed to be infectious may not be [4]. costs for these patients are significant as well. A There is now growing recognition that patients major driver of high healthcare costs and poor out- with idiopathic encephalitis may have treatable comes in encephalitis cases is the fact that more conditions if rapidly and accurately diagnosed. than 50% remain undiagnosed despite often exhaustive, expensive, and invasive diagnostic test- ing [1]. The present review highlights recent prog- ress in two fields, metagenomics and proteomics, Department of Neurology, University of California San Francisco, San that will likely improve the diagnosis rate in cases Francisco, California, USA of encephalitis. Correspondence to Michael R. Wilson, MD, UCSF Multiple Sclerosis Diagnosis of infectious cases of encephalitis was and Neuroinflammation Center, 1500 Owens Street, Suite 320, San revolutionized in the 1990s with the development Francisco, CA 94158, USA. Tel: +415 353 2069; e-mail: michael. of PCR-based assays by Saiki et al. [2]. The ability to [email protected] rapidly and sensitively diagnosis herpes simplex Curr Opin Neurol 2015, 28:283–287 virus (HSV) encephalitis with PCR coupled with DOI:10.1097/WCO.0000000000000198 1350-7540 Copyright ß 2015 Wolters Kluwer Health, Inc. All rights reserved. www.co-neurology.com Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved. Inflammatory diseases and infection next-generation sequencing (NGS), and several plat- KEY POINTS forms are available commercially [8&]. The wealth of Dramatic progress has been made recently in the data generated by NGS have necessitated the devel- speed, cost, and output of new gene sequencers, opment of powerful sequence-based computational referred to as next-generation sequencing (NGS). algorithms to cope with the problem of ever enlarg- ing datasets. In response to this demand, the field of NGS has enabled the application of metagenomics to bioinformatics has expanded in parallel, allowing the discovery of pathogens in infectious encephalitis cases, with significant diagnostic, public health for rapid analysis of large and complex datasets. monitoring, and treatment implications. Given the enormity of the data generated in a typical sequencing experiment, a cloud-compatible Proteomics is the study of proteins recovered directly bioinformatics pipeline has been developed to from the environment. leverage existing filter and alignment algorithms & Proteomics-based techniques are particularly useful for to improve processing speeds [9 ]. Instead of days the identification and further study of novel to weeks, metagenomic data can now be accurately autoantigens in autoimmune encephalitis. processed in clinically actionable timeframes of minutes to hours. For patients with encephalitis, metagenomic INFECTIOUS ENCEPHALITIS data are usually obtained from cerebrospinal fluid The history of encephalitis diagnostics is rich in (CSF) or brain biopsy tissue. First the raw genetic molecular techniques. In addition to virus-specific sequence data, typically 10–20 million individual PCRs, PCR amplification of 16S rRNA and 28S rRNA sequences, each 50–200 nucleotides long, are is well established for identifying bacterial and fun- filtered to remove low-quality sequences and any gal pathogens, respectively [5]. Multiplex PCR assays sequences that align to the human genome. This have also been developed to rapidly test for the step typically removes 98–99% of the raw sequen- presence of multiple neuroinvasive pathogens [6]. ces. Thus, the remaining 1–2% of the sequence In one example called MassTag PCR, conventional fragments, typically numbering in the thousands PCR primers are labeled with molecules of known or hundreds of thousands, are likely to be non- masses called Masscodes. If DNA matching any of human in origin. Their provenance is determined the primers is present in a sample, it will be ampli- by searching for similarities to all the genomic data fied. After this step, Masscode-labeled primers contained in GenBank, the NIH genetic sequence remain bound to the sample. Unbound primers database curated by the National Center for Bio- are washed away, and the Masscodes remaining technology Information (http://www.ncbi.nlm.nih. on the bound primers are identified with mass gov). GenBank now contains over one trillion spectrometry (MS). MassTag PCR has been validated sequences from all manner of living organisms and is now commercialized as FilmArray, a test that (http://www.ncbi.nlm.nih.gov/genbank/statistics). screens a sample for 17 meningitis and encephalitis By searching GenBank, one is able to interrogate a causing pathogens [7]. Although FilmArray is low patient sample for the presence of any type of infec- cost, efficient, and highly specific, the primary tious agent (e.g., fungi, viruses, bacteria, and para- limitation is the requirement for sequence-specific sites) except for prions as the latter lack nucleic primers and thus knowledge of each microbe’s acids. In addition, the search algorithms are flexible genomic sequence prior to testing. Although the enough that they register both exact matches and FilmArray is a multiplex assay, it still functions sequences that may only be 20% similar to a pre- fundamentally as a traditional candidate-based viously deposited sequence [10]. Thus, there is the diagnostic regime that combines separate assays, ability to identify novel pathogens that may only each developed to identify individual pathogens have a distant phylogenetic relationship to a known of interest. microbe. This latter fact is a critical advance for Metagenomics-based approaches radically improving public health surveillance against the depart from traditional candidate-based infectious wide range of novel microbes, many of which are disease diagnostics in the sense that they simul- neuroinvasive viruses [11,12]. taneously interrogate all of the genetic material NGS is an important step forward from the (i.e., host and nonhost RNA and DNA) recovered traditional candidate-based approach to infectious from a biologic sample in an unbiased fashion. disease diagnostics in which each individual diag- Metagenomics is made possible by the dramatic nostic test (e.g., antibody test, pathogen-specific progress that has been made in the speed, cost, PCR assay) is tailored to a particular infection and output of new gene sequencers. Collectively, suspected by the physician based on a patient’s current sequencing technologies are referred to as medical history, risk factors, and physical exam. 284 www.co-neurology.com Volume 28 Number 3 June 2015 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved. A tale of two approaches Schubert and Wilson The unbiased approach offered by NGS has three immunosuppressed post allogeneic bone marrow potential advantages: identification of an entirely transplantation. His neurologic condition progressed novel microbe for which a traditional candidate- over several months to include additional cranial based test does not exist, identification of a known nerve deficits, stupor, and ultimately coma. MRI microbe that is not known to cause a particular revealed injury to deep grey matter and brainstem patient’s disease phenotype, and identification of structures, a pattern consistent with viral injury. A a microbe known to cause a patient’s disease phe- comprehensive clinical evaluation was negative for notype that is nevertheless rarely tested for because